Modern search engines have revolutionized retrieval of unstructured (i.e. textual) information from large data repositories. However, the retrieval of structured information from databases is often limited by traditional approaches that are tightly integrated with a particular relational database design.
Searches on structured data are usually performed on databases. Specialized retrieval software comes into play when answering a query involves searches on both texts and structured data or when queries need to be answered against structured data from specific application domains. Specialized retrieval software is also suitable for answering queries about structured data in cases where there are specific performance requirements or where the operations are performed in novel or unusual landscape configurations.
Structured data stored in relational database tables can be used to model highly constrained objects in specialized application domains in such a way that sophisticated processing is required to extract that data and make it available for business use. A retrieval engine that is optimized to work with such specialized data models can offer levels of performance and flexibility that greatly exceed those achieved with existing general-purpose databases.
To facilitate the development of fast and flexible retrieval services for structured data in the context of modern business applications, what is needed is a logically powerful metamodel that not only supports the modeling of objects that are important in those business applications but also allows the development of efficient implementations of those retrieval services.